2 from __future__ import absolute_import
3 import glob, mmap, os, struct
5 from bup import _helpers
6 from bup.compat import range
7 from bup.helpers import log, mmap_read
12 extract_bits = _helpers.extract_bits
18 """Wrapper which contains data from multiple index files.
19 Multiple index (.midx) files constitute a wrapper around index (.idx) files
20 and make it possible for bup to expand Git's indexing capabilities to vast
23 def __init__(self, filename):
25 self.force_keep = False
27 assert(filename.endswith('.midx'))
28 self.map = mmap_read(open(filename))
29 if str(self.map[0:4]) != 'MIDX':
30 log('Warning: skipping: invalid MIDX header in %r\n' % filename)
31 self.force_keep = True
32 return self._init_failed()
33 ver = struct.unpack('!I', self.map[4:8])[0]
34 if ver < MIDX_VERSION:
35 log('Warning: ignoring old-style (v%d) midx %r\n'
37 self.force_keep = False # old stuff is boring
38 return self._init_failed()
39 if ver > MIDX_VERSION:
40 log('Warning: ignoring too-new (v%d) midx %r\n'
42 self.force_keep = True # new stuff is exciting
43 return self._init_failed()
45 self.bits = _helpers.firstword(self.map[8:12])
46 self.entries = 2**self.bits
47 self.fanout = buffer(self.map, 12, self.entries*4)
48 self.sha_ofs = 12 + self.entries*4
49 self.nsha = nsha = self._fanget(self.entries-1)
50 self.shatable = buffer(self.map, self.sha_ofs, nsha*20)
51 self.which_ofs = self.sha_ofs + 20*nsha
52 self.whichlist = buffer(self.map, self.which_ofs, nsha*4)
53 self.idxnames = str(self.map[self.which_ofs + 4*nsha:]).split('\0')
58 def _init_failed(self):
61 self.fanout = b'\0\0\0\0'
62 self.shatable = b'\0' * 20
67 s = self.fanout[start:start+4]
68 return _helpers.firstword(s)
71 return str(self.shatable[i*20:(i+1)*20])
73 def _get_idx_i(self, i):
74 return struct.unpack('!I', self.whichlist[i*4:(i+1)*4])[0]
76 def _get_idxname(self, i):
77 return self.idxnames[self._get_idx_i(i)]
80 if self.map is not None:
84 def exists(self, hash, want_source=False):
85 """Return nonempty if the object exists in the index files."""
86 global _total_searches, _total_steps
89 el = extract_bits(want, self.bits)
91 start = self._fanget(el-1)
92 startv = el << (32-self.bits)
96 end = self._fanget(el)
97 endv = (el+1) << (32-self.bits)
98 _total_steps += 1 # lookup table is a step
99 hashv = _helpers.firstword(hash)
100 #print '(%08x) %08x %08x %08x' % (extract_bits(want, 32), startv, hashv, endv)
103 #print '! %08x %08x %08x %d - %d' % (startv, hashv, endv, start, end)
104 mid = start + (hashv-startv)*(end-start-1)/(endv-startv)
105 #print ' %08x %08x %08x %d %d %d' % (startv, hashv, endv, start, mid, end)
107 #print ' %08x' % self._num(v)
110 startv = _helpers.firstword(v)
113 endv = _helpers.firstword(v)
115 return want_source and self._get_idxname(mid) or True
119 count = self._fanget(self.entries-1)
120 for ofs in range(0, count * 20, 20):
121 yield self.shatable[ofs : ofs + 20]
124 return int(self._fanget(self.entries-1))
127 def clear_midxes(dir=None):
128 for midx in glob.glob(os.path.join(dir, '*.midx')):